Dezhi Dai

Dezhi Dai

Auto-generated profile, based on publicly available information

Dezhi Dai

Argonne National Laboratory (ANL), Argonne, IL (United States)

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 May 2019 in International Journal for Numerical Methods in Fluids
Verified
Journal Article
Closed access
PDF file
DDDezhi Dai
Dezhi Dai
ATAlbert Y. Tong
Albert Y. Tong

30 July 2018 in International Journal for Numerical Methods in Fluids
Verified
Journal Article
Closed access
PDF file
DDDezhi Dai
Dezhi Dai
ATAlbert Y. Tong
Albert Y. Tong

28 October 2025 in Nuclear Science and Engineering
Verified
Journal Article
Closed access
DDDezhi Dai
Dezhi Dai
EMElia Merzari
Elia Merzari
LBLandon Brockmeyer
Landon Brockmeyer
DSDillon Shaver
Dillon Shaver

30 September 2026
Verified
Report
Closed access
PDF file
A(Argonne National Laboratory (Anl)
Argonne National Laboratory (Anl)
HYHaomin Yuan
Haomin Yuan
JFJ. Fang
J. Fang
DDD. Dai
D. Dai
DSD. Shaver
D. Shaver
GGGuilherme Gottems
Guilherme Gottems

30 September 2025
Verified
Report
Open access
PDF file
HYHaomin Yuan
Haomin Yuan
A(Argonne Argonne National Laboratory (Anl)
Argonne Argonne National Laboratory (Anl)
NSNadish Saini
Nadish Saini
DDDezhi Dai
Dezhi Dai
DSDillon Shaver
Dillon Shaver
NHNathaniel Hoyt
Nathaniel Hoyt

22 July 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
HYHaomin Yuan
Haomin Yuan
DDDezhi Dai
Dezhi Dai
DSDillon Shaver
Dillon Shaver
YLYuhsiang Lan
Yuhsiang Lan
GGGuilherme Gottems
Guilherme Gottems
EMElia Merzari
Elia Merzari
Accurate prediction of thermal radiation remains a major challenge in high-fidelity CFD simulations, particularly for large-scale problems solved using high-order methods on modern GPU architectures. While classical radiation models such as the P1 approximation and surface-to-surface (view-factor) methods are well established, their efficient integration into scalable, high-order CFD frameworks remains limited, especially for complex geometries and large numbers of radiating surfaces.In this work, a scalable thermal radiation modeling framework is developed within the GPU-accelerated spectral element code NekRS. Two complementary approaches are implemented: a P1 formulation for participating media, directly coupled to the energy equation through a radiation source term, and a surface-to-surface model for non-participating media based on precomputed view factors. To enable large-scale applications, a GPU-accelerated view-factor calculator is developed using a hemi-cube rasterization strategy, combined with an element-wise, matrix-free formulation that avoids explicit storage of dense view-factor matrices and significantly reduces memory requirements.The proposed framework is systematically verified against canonical benchmarks with analytical solutions, including parallel plates and concentric spheres, and further validated through cross-code comparisons with established CFD solvers. Excellent agreement is observed across all cases. Additional simulations in laminar channel flow, pebble-bed configurations, and natural-convection cavities demonstrate the capability of the method to capture coupled convection–radiation heat transfer in complex geometries. In the pebble-bed case, radiation is shown to contribute up to approximately 40\% of the total heat transfer, highlighting its critical role in high-temperature applications.Overall, this work provides a scalable and high-fidelity pathway for incorporating thermal radiation into advanced CFD simulations, bridging the gap between accurate flow modeling and predictive radiative heat transfer.